Skip Navigation LinksHome / Research Background

Comorbidity of Depression and Generalized Anxiety Disorder (June 20-22, 2007) 

Prepared by Michael B. First, M.D., DSM Consultant to the American Psychiatric Institute for Research and Education (APIRE), a subsidiary of the American Psychiatric Association

The APA, in collaboration with the WHO and NIH, convened a diagnosis-related research planning conference focusing on depression and generalized anxiety disorder at the Institute of Psychiatry in London on June 20-22, 2007. The conference was the tenth in a series of 12 NIH-funded conferences on "The Future of Psychiatric Diagnosis: Refining the Research Agenda" that is administered by APA's American Psychiatric Institute for Research and Education (APIRE). The conference co-chairs were Kenneth S. Kendler, MD, Virginia Institute for Psychiatric and Behavioral Generics at Virginia Commonwealth University (Richmond, VA) and David Goldberg, MD, Institute of Psychiatry (London, UK). Twenty three invited scientists from around the world participated.

Kenneth Kendler MD, PhD (Richmond, VA) began the conference by providing an overview of the problem that the conference was set up to address, namely to clarify the diagnostic and nosological relationship between Major Depressive Disorder (MDD) and Generalized Anxiety Disorder (GAD), i.e., whether they are different forms of the same disorder, closely related disorders, or distally-related disorders. MDD and GAD will be compared with respect to a range of available validators, including genetics, biology, treatment, development, course predictors, disability, and psychosocial stressors. Dr. Kendler then raised the question of how do we weigh different classes of validators when they do not agree. This issue was confronted in the 1980's when looking at subtypes of schizophrenia. Using outcome as a validator, some definitions did much better but would do worse on validators such as temporal stability and family history. This requires the development of a hierarchy of validators, a task which is not inherently an empirical question, i.e., should validators be chosen that most closely reflect underlying etiology like genetic factors or should more pragmatic validators, like treatment response and course, be given precedence. Given that this reflects a system-wide issue, is it possible to establish a deeper set of principles that might be applied more generically to other areas of DSM-V?

Two major approaches can be used to guide the issue of placement of disorders within the DSM; a descriptive approach, in which disorders are grouped together according to surface features or clinical characteristics or a causal approach based on the underlying etiology. DSM has strongly taken a descriptivist approach; is it now time to move toward an etiologically-based approach? Dr. Kendler also noted that models used to consider the MDD/GAD issue have focused largely on shared risk factors. What should be done in situations in which disorders have direct causal relationship, for example, that having GAD directly increases risk for MDD? Finally, Dr. Kendler wondered how to avoid the "house of cards" phenomenon caused by the inter-relationship of disorders in the DSM. How can we consider moving MDD without evaluating the relationship between MDD and bipolar disorder? Furthermore, if GAD is closely related to MDD, what about other anxiety disorders like panic or social phobia which have also been shown to have high comorbidity with MDD? Dr. Kendler concluded by noting that nosological decisions are like setting social policy given that they inevitably go beyond empirical data to deeper conceptual issues about the nature of psychiatric diagnosis.

Jack Hettema, MD PhD (Richmond, VA) then gave a presentation comparing GAD and MDD on a variety of antecedent, concurrent, and predictive external validators in order to determine whether GAD can stand as a valid diagnosis on its own, distinct and separable from MDD and whether this overlap is unique to GAD or also applies to the other anxiety disorders. In terms of genetic risk, family, twin, and high-risk transmission studies all indicate that GAD and MDD share some, if not most, of their genetic risk factors. However, similar sources of data suggest that other anxiety disorders, particularly panic disorder, may share some genetic risk with both GAD and MDD, although perhaps not to the degree their share risk with each other. In terms of childhood environmental risk, studies examining measures of parental loss, parenting style, and abuse converge on findings of modest associations of each of these predictors with both adult MDD and GAD. Dr. Hettema noted, however, that these findings are generally non-specific, as these occur in the larger context of similar associations between these risk factors and other anxiety disorders, both individually and as a group. Regarding the impact of stressful life events, they have been consistently implicated in the risk for MDD, and, to a lesser extent, GAD or GAD-like syndromes although there appears to be only partial overlap between the types of events that predispose to the two conditions (i.e., loss predisposes to depression, danger to anxiety). A personality trait of negative affect (e.g., neuroticism from the five factor model of personality traits) seems to function as a general risk factor for MDD, GAD, and possibly other anxiety disorders, accounting for a substantial portion of their comorbidity in several studies. Findings are less consistent regarding the comorbidity of personality disorders with MDD and GAD. Demographic variables (i.e., being female, mid-adult aged, widowed/separated/divorced, and low income or unemployment) have been positively associated with illness in epidemiological samples regardless of whether the diagnosis was MDD, GAD, or panic disorder, suggesting that these may be nonspecific indices of internalizing disorders. Regarding biological validators, there are similarities between MDD and anxiety disorders in terms of decreased heart-rate but differences in terms of measures of attentional bias to emotional stimuli. startle-reflex abnormalities, HPA axis functioning, and sleep abnormalities. Dr. Hettema concluded by noting that although the data supports a strong overlap between GAD and MDD based upon familial/genetic, childhood environment, personality trait, and demographic data, for many of these correlates, the relationship to MDD overlaps substantially with other anxiety disorders, suggesting that the association is somewhat nonspecific. In contrast, data from studies examining life events, personality disorders, biology, comorbidity, and pharmacology are mixed, showing some areas of similarity between GAD and MDD but some clear differences as well. Overall, Dr. Hettema noted that there appears to be little more reason to question the nosologic validity of GAD than that of some of the other anxiety disorders. In his discussion, Jules Angst, MD (Zurich, Switzerland) commented that bipolar disorder should be included in the comparison along with MDD, given that there is no clear distinction between MDD and Bipolar disorder. He noted that most large community studies reported significant associations of GAD with MDD based on life time prevalence rates across all ages and that this is a serious methodological problem which can create pseudo-comorbidity between MDD and GAD. In his longitudinal Zurich Study, where all subjects were of the same age, the association of GAD, panic disorder, and panic attacks with major depressive disorder was lower than the association of these disorders with bipolar disorders. He concluded by noting that his data fail to provide any rational evidence for merging GAD with MDD.

Charles Nemeroff. MD, PhD (Atlanta, GA) presented next on the commonalities and distinguishing factors in the biology of depression and generalized anxiety disorder. He noted that normal anxiety and sadness involve unique brain circuits and that neuroanatomical and functional disruptions differ between patients with GAD and MDD both at baseline, with induction of depressed or anxious moods, and in response to treatment. For example, acute sadness activates the dorsal insula whereas acute anxiety activates the ventral insula, and the amygdala is overactive at rest in primary mood disorders whereas it is overactive only during symptom provocation in anxiety disorders. From a neuroendocrinological perspective, depression is associated with a number of abnormalities (blunted TSH response to TRH , dexamethasone non-suppression, elevated plasma ACTH and cortisol levels, elevated CSF CRF concentrations, all, except the TSH response normalizes after successful treatment); these findings have not been observed in patients with GAD (e.g., baseline cortisol levels in young and elderly patients with GAD do not differ from age-matched healthy control subjects, CSF CRF levels are not elevated, etc). Furthermore, there is increasing evidence for the role of dopamine in depression (e.g., dopamine neurons in VTA mediate reward systems) but no evidence of any dopamine abnormality in anxiety disorders. Finally, MDD and GAD patients exhibit distinct sleep disturbances (e.g., shortened REM latency in MDD but not GAD). Dr. Nemeroff concluded by arguing that based on neuroanatomical, neuroendocrinological, and sleep polysomnographic findings, MDD and GAD are not the same disorder although he noted that the validity of some of his points depended on the ability to extrapolate from studies of symptoms of sadness to the diagnosis of MDD and symptoms of anxiety to the diagnosis of GAD.

Toshiaki Furukawa MD, PhD (Nagoya, Japan) presented data on the similarities and differences in response to treatment between MDD and GAD. Given that three classes of psychotropic medications, namely antidepressants, benzodiazepines and azapirones, have been suggested to be effective for GAD, are they differentially effective for MDD? To answer this question, Dr. Furukawa did a systematic review of randomized controlled trials (RCT) of these types of medications for the acute treatment of MDD and GAD using as outcome measures the HAM-D for depression and the HAM-A for anxiety. He found that by pooling effect sizes for the studies, antidepressants, benzodiazipines and azapirones are better than placebo for treatment of both anxiety and depression in both GAD and MDD with similar effect sizes. He proposed a number of possible explanations for these findings, including 1) GAD and MDD are the same disease entity, 2) GAD is almost always accompanied by subsyndromal depression, and MDD by subsyndromal general anxiety; 3) current outcome measures of anxiety and depression are inappropriate and/or insensitive; 4) these psychopathological concepts are confounded; 5) study publication or outcome reporting bias may have overestimated the treatment effects; and 6) any psychoactive substance can beat placebo if the problem of statistical power is overcome. He suggested that based on these results, if there remains any phenomenological or pathophysiological specificities to GAD, an alternative would be to consider it a subtype of MDD (i.e., "with generalized anxiety") and proposed a reanalysis of all GAD and MDD trials which had measured HAM-A and HAM-D at the same time, possibly focusing on more depression-specific and anxiety-specific subscales.

In his discussion David Kupfer MD (Pittsburgh, PA) commented on the quality of the data and sensitivity of the methods used in various studies that were reviewed in the prior two presentations. He noted that information about "behavioral phenotype" for depression and anxiety should go beyond the diagnostic criteria. Furthermore, a contemporary adaptation of the Robins and Guze validators for DSM-V needs to include the neurobiological profile (e.g., susceptibility genes, pharmacogenomics, etc.), treatment response and context/environment. He noted that because the treatment response presentation employed meta-analytic techniques, it reduced the number of available studies. On the other hand, the rigor of the analysis could have been improved if the newer tools of evidence-based medicine were employed. Furthermore, future treatment dissection studies should also use psychosocial interventions which may be cleaner and more useful.

David Goldberg, MD (London, UK) presented on the psychometric structure of depression and anxiety, beginning with research measuring symptoms using standardized tests. Negative affect tends to appear as a large distress factor cutting across many different scales. If measured during health, "negative affect", is a good trait measure that assesses the common variance between common mental disorders but, measured during an episode of illness, the trait variance is masked by state variance. In addition to high negative affect, depression has low positive affect which is a bipolar dimension that runs from loss of interest or pleasure to psychomotor retardation, apathy, extreme fatigue and lethargy, and shows both diurnal and seasonal fluctuations. Some investigators have offered solutions to the state-trait problem by proposing that there are two underlying systems: a behavioral activation system (BAS) (approach-related, positive incentive motivation); and a behavioral inhibition system (BIS) (sensitivity to threat and non-reward). The behavioral activation system is stable with varied clinical states, and treatment only affects the behavioral inhibition system. Results from structured interviews in both community and clinical settings show correlations between scales measuring anxiety and depression often well above 0.5. Dr. Goldberg attributes the high levels of reported comorbidity between depression and anxiety to the high common component of negative affect. Research focussing on clinical diagnoses of mood and anxiety disorders has shown that the internalizing disorders can be divided into two big groups: distress syndromes and fear syndromes. Twin studies indicate that GAD and MDD have the same additive genetic factors, that neither is affected by shared environment, and that unique environmental factors are important for both. GAD in longitudinal data sets occurs in three situations: a stage in the development of major depression, a stage in recovery from major depression, and an illness occurring independently. Linking categorical and dimensional data, autonomic arousal relates only to panic and agoraphobia, all internalizing disorders are affected by negative affect, and low positive affect relates only to MDD and social phobia. Dr. Goldberg concluded by noting that although categories continue to be necessary (i.e., clinicians need to decide whether or not to treat), they should be supplemented with dimensions. Furthermore, it follows from the data that MDD and GAD are not separate disorder or the same disorder but closely related disorders, and that MDD and GAD should be placed along with the other mood disorders in a diagnostic grouping called "dysphoric disorders" with the remaining anxiety disorders placed in a grouping called "fear disorders." In his discussion, Patrick Shrout, PhD (New York, NY) noted that many of the studies reviewed by Dr. Goldberg are self-report in which respondents are presented with a task that has a high cognitive load (i.e., recalling a period vividly, evaluating the number of days in that period, and evaluating the severity and extent of symptoms) and high social demands (i.e., help seeking goals, conversational norms in epidemiological interviews). These demands can lead to biases that bleed across disorders and result in artificially high correlations. Dr. Shrout also noted that although there is an important emphasis on traits (i.e., neuroticism, negative and positive affect), state variation (which includes response to therapy) is also important and is often confounded with measurement error and is minimized by cognitive processes. Dr. Shrout recommended that efforts be made to obtain time intensive basic data, i.e., basic percepts of mood and functioning on a daily basis, analogous to obtaining ambulatory blood pressure measurements in primary care studies of hypertension. The variance could then be decomposed into stable trait-like variation, systematic developmental variation, systematic state variation, and nonsystematic (error) variation. Dr. Shrout concluded by noting that when people use multi-variate methods to make strong claims, it is not definitive and one must examine the robustness of the claims across models. In his discussion, Timothy Brown, PsyD (Boston, MA) summarized data on the boundary between GAD and MDD from patient samples in his center in Boston. He concluded that GAD and MDD are closely related disorders that should not be combined. He also noted that MDD and GAD are highly comorbid (e.g., 74% of individuals with GAD have had a lifetime mood disorder), the actual extent of which is masked by the DSM hierarchy rules. When looking at structural analysis, the best fitting measurement model kept GAD and MDD as distinct latent variables. Features that distinguish between GAD and MDD include MDD having a much stronger relationship with positive affect/behavioral activation (being more similar to social phobia in this regard), the lack of a relationship between MDD and autonomic arousal (in contrast with GAD's inverse relationship), the fact that cross-sectional overlap is not excessive (r=0.49-0.63), modest temporal co-variation, and temporal sequencing (i.e., in comorbid cases, MDD and GAD onsets differ by 7+ years). Dr. Brown concluded by recommending that movement toward a formal diagnostic system with dimensional elements could begin with the use of dimensional ratings of the current DSM-IV criteria sets and that evidence from clinical samples does not support any current proposals for re-organization of anxiety and mood disorders (e.g., fear disorders, distress disorders) because of the degree of overlap between individual diagnoses and major categories of autonomic arousal, negative affect, and positive affect.

Ronald Kessler, PhD (Boston, MA) presented next on the relationship of GAD and MDD over time in the US National Comorbidity Survey Panel sample (NCS-2). In this community survey conducted from 1990 to 1992 in the United States, 8000 people were interviewed and 5000 were followed up 10 years later. Dr. Kessler noted that according to the NCS data, there is nothing specific about the relationship between non-bipolar depression and GAD; people with mania and depressive episodes have the same high comorbidity. Lifetime prevalence rates were 28.8% for MDD vs. 11.9% for GAD with 9.6% having a first onset of MDD during the 10 year follow-up period vs. 3.6% for GAD. However, if a one-month duration is used for GAD (as opposed to the current 6 month duration), the prevalence of GAD is very similar; in the community there are many people who meet GAD criteria for only one month but who do not present for treatment-only those who get depressed show up for treatment. Considering the conditional prevalence of MDD and GAD (i.e., the percentage of those with GAD who have subsequent MDD and vice versa), having GAD or MDD increases the likelihood of developing the other in the future, with the likelihood of GAD predicting the subsequent onset of MDD higher than MDD predicting subsequent GAD. A wide range of childhood adversities predict later onset of MDD and GAD, with little difference between the two disorders. Regarding persistence of the disorders, of the people who at baseline had a lifetime history of MDE, 50% had an episode in the next decade; for GAD, 75% had an episode in the next decade. In terms of what predicts persistence, the younger you are when GAD starts, the more persistent it is over the lifespan. In predicting persistence of MDE, comorbid GAD is a significant predictor. Dr. Kessler concluded by summarizing that the NCS data supports the hypothesis that MDD and GAD are closely related conditions but that it is worth noting that other disorders are also highly correlated with GAD and MDD; panic disorder in particular is highly correlated with both MDD and GAD and dysthymic disorder is even more highly correlated with GAD than is MDD (although if the duration of GAD is reduced to only one month, then MDD is more highly correlated). The differences in course between MDD and GAD indicate that despite the overlap, there is still something separate going on. Looking cross-nationally, there is substantial variation across countries and within countries; there is a 2:1 gender difference in depression across countries but the gender gap is closing. Dr. Kessler noted that the traditionality of sex role orientation is a powerful predictor of gender differences; as roles get more equal, the difference gets smaller. In his discussion, Valery Krasnov, MD (Moscow, Russia) described a program for the recognition and treatment of affective spectrum disorders in primary care (i.e., in 5000 outpatients screened; affective spectrum disorders were recognized in 61.2% of primary care patients; 35.8% were evaluated by psychiatrists and 30.1% patients were considered to have depression, 24.8% were advised to have treatment with antidepressants, and 14.2% completed the 6 week course of treatment). Only 4.4% of the sample could be diagnosed as having a "pure" (i.e., non-comorbid) anxiety disorder. Dr. Krasnov proposed that there are three stages of depression that could be recognized in this sample: a prodrome stage with non-specific symptoms of emotional and vegetative instability, an anxious stage which included three sub-stages (situational anxiety with a specific cause, free-floating anxiety, and anxiety without cause), and a depressive stage with three sub-stages (depression with anxious foreground, depression with hidden anxiety, and depression with hyporeactivity and psychomotor retardation). He concluded that the frequent combination of anxiety and depression but only rare cases of pure anxiety or depression in the primary care population are in conflict with the concept of separate comorbid disorders. In his discussion, David Fergusson, PhD (Christchurch, New Zealand) commented that the evidence strongly suggests overlapping disorders with some specificity of symptoms, causes, and longer term outcomes. He noted that Dr. Kessler raised the conjecture that after adjustment for common causes, GAD may influence MDD and MDD influence GAD. Using data from the Christchurch Developmental Study (CHDS) at three time points, Dr. Fergusson devised two models that included both common causes and reciprocal effects, with the results suggesting that exposure to MDD increases risk of GAD but exposure to GAD does not increase MDD. He noted that both studies suggest causal relationships between MDD in GAD over and above the effects of common causes but that they differed in terms of Kessler finding a reciprocal effect; CHDS suggests a unidirectional effect. He concludes that internalizing symptoms can be partitioned into components reflecting both a generalized tendency to internalizing symptomatology and disorder-specific components.

K.S. Jacob, MD (Vellore, India) presented on the cross-cultural aspects of anxiety and depression, reporting on a factor analysis of anxiety and depression done cross-culturally. Exploratory factor analysis (EFA) simplifies the description of complex phenomena by identifying unifying constructs that parsimoniously describe correlations among a related group of variables (latent traits). Such analyses have led to two factor models (anxiety-depression) but these factors have been highly correlated. Professor Jacob noted a number of criticisms of exploratory factor analysis, namely that there is subjectivity in comparison of factor structure, that the technique is not hypothesis driven and that the method is sensitive to psychometric properties of the scales employed. For this EFA, data sets were used from 11 different settings (from seven countries) using over 19,000 patients. He found that many of the data sets had 2 factor anxiety-depression solutions whereas some had 3 or 4 factor solutions with somatic symptoms, panic and phobia being the main loadings. Other confirmatory factor analyses were used to check one factor vs. two factor models, the results being that when models are made equal across settings and culture, the models come out poorly. There were wide variations for phobia, panic, obsessions, and compulsions, particularly for the two factor model for sites in South Asia (India, Pakistan) and Africa (Ethiopia and Tanzania). Professor Jacob summarizes his findings by concluding that one and two factor models have similar fit, with the two factor model being very slightly superior. Overall, the same model does not fit well across cultures. He argued that the one factor model is more parsimonious and reflects a general distress factor in primary care and community settings.

There were three discussants for this presentation. Gerhard Heinze MD (Mexico City, Mexico) noted that the close association between GAD and MDD may be explained in a number of ways. It may indicate that both clinical entities are distinct forms of the same underlying disorder with comorbidity being an index of disease severity, or that MDD and GAD may be separate disorders deriving from the same underlying diagnostic category, or they may be independent disorders linked by a common pathogenic pathway in which case shared symptoms are an expression of these pathogenic factors Dr. Heinze commented that the study by Dr. Jacob offers an ample source of culturally and ethnically diverse countries. However, by using primary care or community surveys, the variables tend to be very skewed and influenced by prevalence which may make the factor analysis very unstable. Although the one-factor solution is a more parsimonious explanation, it does not necessarily mean it is the best one. Given that factorial analysis is insensitive to specific time patterns of comorbidity (e.g., GAD as a risk factor for MDD), the results may be problematic to interpret. Ultimately, based on Dr. Jacob's paper, the question of whether MDD may be a part of the anxiety disorders or GAD being part of the mood disorders is difficult to confirm.

In his discussion, Vikram Patel.PhD (London, UK) commenting on the findings of the Jacob presentation, noted that a one factor solution with depression and generalized anxiety fits almost as well as a two factor solution, but there is considerable heterogeneity between data sets from the various sites and countries on the symptom loadings. However, even in a two factor solution, the second anxiety factor comprises mainly panic, phobia and obsessions and compulsions and not GAD. In fact "co-morbidity" of MDD and GAD may really be a euphemism for two sides of the same coin. In the cross-cultural context where more than 95% patients with depression and anxiety consult in primary care settings only, mixed presentations are overwhelmingly dominant. Dr. Patel also pointed out the massive overlap of depression and anxiety with somatoform disorders, which may all fit under the larger rubric of dysphoric disorders. Summarizing some of the conclusions from the Somatoform Presentations research planning conference (which he co-chaired), emotional and unexplained somatic symptoms cannot be easily separated into specific clusters and discrete somatoform diagnoses are rare. Latent trait and cluster analysis do not support separate DSM-IV discrete categories and instead emotional and somatoform diagnoses may load into one common factor ('internalizing' disorders).

Dan J Stein, MD PhD (Cape Town, South Africa), in his discussion, reviewed the findings of depression and anxiety across cultures, suggesting three potential approaches. The first, the clinical (positivist) approach, views science as describing universal laws and sees psychiatric classification as being based on universal principles. The second approach, the anthropological (hermeneutic) approach, views psychiatric classification itself as a cultural product and stresses that the expression and experience of illness varies from culture to culture. A third approach synthesizes the first two approaches, viewing science is a social activity, but one that discovers real mechanisms. It distinguishes between disease (psychobiological mechanisms) versus illness (experience of disease), noting that some disease mechanisms are universal, but that illness varies across time and culture. Although MDD and GAD involve universal processes, culture and society affect their diagnosis and treatment, expression, and experience. The argument that a single construct of depression/anxiety useful for understanding the experience of most patients in primary care may have some strengths, but there is also work on the psychobiology and treatment of depression/anxiety disorders that suggests the value of continuing to differentiate these conditions. Dr. Stein concluded by noting that further research is needed on the increased risk of female gender and lower socioeconomic status for MDD/GAD and that delineating disease and illness is useful even in patients within Western cultures.

The next set of presentations focused on early predictors of GAD and MDD. The first presenter, Terrie Moffitt, PhD (London, UK) analyzed prospective longitudinal data from the Dunedin study (a 1972-73 New Zealand cohort followed longitudinally, last assessment at age 32) to help answer three questions: 1) is GAD the prodrome to MDD, i.e., does GAD always precede MDD and do all GAD cases eventually develop MDD? 2) are childhood risk factors for MDD and GAD the same, or different? and 3) is it true that GAD is always the milder disorder? Regarding the prevalence of MDD by age and sex (from ages 11 to 32), as one would expect there is a higher prevalence in women, with a sharp increase in prevalence at age 16-18; the further along the cohort is followed, the more recurrent cases of MDD appear. The prevalence of GAD (using DSM-III criteria with a one month duration) was higher than expected (e.g., >15% of women at age 32) All three childhood anxiety disorders (overanxious disorder, childhood phobias, separation anxiety) predict GAD, with childhood phobias predicting GAD better than overanxious disorder. Regarding the answer to the first question (is GAD the prodrome to MDD), GAD precedes MDD 42% of the time whereas MDD precedes GAD 32% of the time, and in 28% of cases they are concurrent. Furthermore, the more years a study subject presented with a diagnosis of GAD, the more likely he or she was to experience depression and vice versa (the more episodes of recurrent depression, the more GAD). Regarding the answer to the second question (i.e., are the risk factors for GAD and MDD the same), most individuals who have had GAD and MDD by age 32 are the same people this would virtually insure similar antecedent correlates. Therefore, groups with pure disorders were isolated for comparison. Comparing risk factors for MDD and GAD (childhood inhibited temperament at age 3-5, childhood maltreatment, childhood internalizing symptoms, mother's internalizing symptoms, informant reported neuroticism, family history of anxiety, family history of depression, and low positive emotionality), risk factors were different for the different groups: the comorbid group in all cases had the highest prevalence of risk factors with the "GAD only" group next. (The exceptions were family history of MDD and low positive emotionality, for which the "MDD only" group had the next highest prevalence after the comorbid group). The sharing of many risk factors for GAD and comorbid GAD+MDD suggests that the presence of GAD may signal a pathway toward more severe disorder. To answer the third question (i.e., is GAD a milder disorder than MDD), service use was examined with the result being that most service use was by people who had both disorders, the "MDD only" and "GAD only" groups were similar and used fewer services than the comorbid group. Overall, it appears the MDD and GAD are more balanced and symmetric than previous thought, in their sequence of onset, risk antecedents, and service use. Dr. Moffitt concluded by noting that these findings question whether GAD comes first and whether MDD is the more serious form. It appears in this longitudinal cohort that GAD and MDD are strongly related but that neither is earlier or milder than the other.

Marcus Richards, PhD (London, UK) then presented data from the MRC National Survey of Health and Development, also known as the British 1946 birth cohort, to explore whether there are early adverse exposures that differentiate depression and anxiety risk. The community version of the Present State Examination (PSE) was given at age 36 years, and depression/anxiety risk was determined using measures and ratings of mental health and temperament in adolescence along with a wide range of potential explanatory variables including indicators of early adversity. PSE factor analysis based on previous work by Drs. Tim Croudace and George Ploubidis found two second-order factors, anxiety and depression, that are extremely highly correlated. Following this, CATEGO-derived PSE cases were disaggregated by Dr.Richards into anxiety only (n = 201), depression only (171), and mixed anxiety and depression (n = 37), the remainder of the cohort serving as a normal comparison group. Early adverse exposure variables (previously identified as PSE caseness prediction variables) included father's age at birth, death of a parent up to 15 years, separation from parents up to 6 years, parental divorce or separation up to 15 years, maternal neuroticism, chronic illness up to 5 years, maternal management at 4 years, parental health at 14 years, and mild learning disability at 15 years. Intermediate phenotypes included self-reported neuroticism at 13 years; and adolescent internalizing disorder based on teacher ratings at 13 and 15 years. Summarizing the results of the regression analyses, several variables differentiated anxiety or depression from normal, especially learning disability and adolescent neuroticism or internalizing disorder, but did not differentiate anxiety from depression. Dr. Richards concluded that, to the extent that the PSE-derived groups represent MDD and GAD, this study does not suggest that these two disorders are only distally related. However, the data cannot distinguish whether they are either different forms of the same disorder, or closely related disorders. In his discussion, Ian Goodyer MD (Cambridge, UK) pointed out that these findings support those from earlier presentations arguing that GAD and MDD are closely related. However, he noted that fundamental differences in time frame between GAD (6 months) vs. MDD (2 weeks) make it hard to compare them. Furthermore, item overlap (irritability being a symptom of depression in children and adolescents and a symptom of adult GAD) also complicate comparisons. He concluded that the results of this study are consistent with other lines of evidence suggesting that anxiety and depression are too closely interlinked to be easily dissociated, whether by descriptive prevalence studies, by risk factor studies, by mathematical modeling, or by intervention studies.

Myrna Weissman, PhD (New York, NY) presented new analyses of data from the Three Generation High Risk Study, a 20 year follow-up of families at high and low risk for depression (high risk vs. low risk defined as MDD or no MDD in generation one (G1)). In this sample, Dr. Weissman found that MDD in G1 does not transmit GAD to generation two (G2) or generation three (G3) whereas MDD in G1 does transmit MDD and comorbid MDD+GAD to G2. Furthermore, GAD in G1 does not transmit MDD to G2 or G3, and MDD-only in G1 increases risk of phobias and other anxiety disorders in G2 and G3. Regarding the sequence of onset over the 20 year follow-up in the second generation, among those with comorbid GAD and MDD, 25% had MDD onset first, 37.5% had GAD onset first, and 37.5% had GAD and MDD onset concurrently. Finally, childhood GAD in G2 does not predict onset of MDD in childhood, adolescence or adulthood in G2 or G3. Summarizing her findings, Dr. Weissman noted that MDD transmits MDD and comorbid MDD and GAD but not GAD by itself, although it does transmit other anxiety disorders, and while prepubertal GAD, a low frequency disorder in this sample does not predict MDD, other anxiety disorders do predict MDD. She concluded that these results suggest that GAD and MDD are only distally related and recommended that GAD and MDD be kept in separate categories but that hierarchies not be used as they are now in DSM-IV.

Martin J. Prince, MD (London, UK) presented on the disability of anxiety and depression in the UK National Psychiatric Morbidity Survey 2000 (NPMS) looking at MDD, GAD, and the ICD-10 category mixed anxiety depressive disorder (MADD). The survey of over 8000 participants collected both symptom counts (using the CIS-R) as well as ICD-10 diagnoses. Dr. Prince found no effect of diagnostic grouping on most outcomes after adjusting for CIS-R symptom score but there was a large independent effect of CIS-R symptom score on all impact outcomes after adjusting for diagnosis, indicating that a full understanding of the public health impact of mental disorders requires assessment of symptom load in addition to diagnostic formulation. MADD accounted for > 50% of all common mental disorders in the sample and had a similar symptom profile to both pure anxiety and pure depression. As expected from its definition, MADD has lower symptom scores than those seen for depressive, anxiety, or comorbid disorders. Despite this, it has similar impact on most outcome indicators as do pure depression and anxiety disorder, although lesser impact than for comorbid depression and anxiety. In fact 20% of all disability days in the UK occur in people with MADD. Dr. Prince remarked that even though the definition of MADD is problematic, it is a disorder associated with impaired function and quality of life and should be operationally defined in ICD-11 and DSM-V. In his discussion, J. (hans) Ormel MD (Groningen, the Netherlands) presented data from the Groningen primary care studies that largely supported Prince's conclusions by confirming the severity gradient that runs from a few generally non-specific distress symptoms, to subthreshold anxiety and/or depressive disorders, to noncomorbid definite disorders, and finally to comorbid definite disorders. The primary care data also suggested that symptom load drives disability, recognition by GP, long-term outcome, etc. Ormel noted that although MADD's public health impact is substantial, its clinical significance is unclear and current definitions in DSM and ICD need improvement. He also summarized the reasons why MDD and GAD should be in the same diagnostic class as related but not identical disorders: they co-occur beyond chance expectations, they share genetic susceptibility and share some risk factors (e.g. neuroticism) but not all, they both belong to the internalizing disorders domain, GAD often precedes MDD, and they respond similarly to SSRI's but differently to benzodiazepines. He emphasized that neurobiological data on the substrate of GAD and MDD are most pertinent and that the adjustment disorders should be taken into account in the MDD-GAD discussion. He concluded by suggesting a diagnostic class of Distress Disorders consisting of (1) Adjustment disorders or problems including subthreshold GAD and depression; (2) MDD; (3) GAD; and (4) Mixed Anxiety and Depression (severe subtype: either MDD+GAD or MDD+ subthreshold GAD or GAD+subthreshold MDD); mild subtype: more severe than adjustment problems but not MDD or GAD).

George Brown, PhD (London, UK), provided a general review of the role of psychosocial factors in the course of DSM-III-R anxiety disorders (including GAD) and depression, along with a discussion of data from a longitudinal study of the impact of psychosocial factors among 404 Islington mothers. He first outlined the role of psychosocial factors in explaining the comorbidity of the two conditions. Over a 3-year period, including two follow-ups, there was a substantial relationship between early neglect and abuse and both DSM-III-R depression and anxiety disorder (except simple phobia and mild agoraphobia). This was sufficient to explain a good deal of their comorbidity in adulthood. Much was due to early maltreatment increasing the prevalence of both conditions in the general population and in so doing greatly increasing their coming together by chance. He then outlined a second possible psychosocial contribution to comorbidity, i.e., the joint occurrence of danger events (tending to provoke anxiety) and loss events (tending to provoke depression) resulting in an increased risk of the onset of both disorders. It is also important to take account of the fact that both the onset and course of depression appear to be much more sensitive to the current environment. For example, a severely threatening life event rarely brings about an onset of depression without the presence of a vulnerability factor such as an ongoing aversive core sexual relationship or low self-esteem. Once an event has occurred there is only a 5% risk of developing depression without one of these but a 50% risk with two. No such evidence has so far emerges for anxiety disorders. Thus, for depression there is evidence for the protective effect of emotional support from a core social tie, but nothing to suggest this holds for anxiety. Although typically triggered by an event conveying 'danger', the onset and course of anxiety disorders appear to be much more encapsulated from the social environment. Dr. Brown concluded by noting that he saw no case for allying GAD more closely with depression, although it is unclear whether having a shorter duration for GAD might change the findings. In his discussion, Sidney Zisook, MD (San Diego, CA) noted that Dr. Brown's findings supported the hypothesis that GAD and MDD are related but how close that relationship is still needs to be elucidated. Dr. Zisook then discussed the relationship between bereavement and depression and generalized anxiety, nothing that while there has been a great deal of work looking at depression occurring in the context of bereavement, there is only one paper that looks at the onset of GAD after bereavement. The risk factor most strongly related to the onset of MDD in late life is recent bereavement, and death of a partner is also a risk factor for anxiety. If one examines the typical biphasic response to loss (i.e., protest, agitation, evolving to despair) it includes not simply depression after loss but also anxiety. Looking at bereavement symptoms post 2 months in widows, all of the symptoms of anxiety were elevated compared to a married comparison group, and they stayed elevated over two years; depressive symptoms decreased over time more than anxiety symptoms. Dr. Zisook concluded his remarks with a look at data from the STAR*D study on the treatment of depression, noting that anxious depression was the norm and was far worse than non-anxious depression in terms of disability, suicidality, and being less likely to respond to treatment. Furthermore, the children of mothers who were depressed in the study had high rates of both depression and anxiety disorders which persisted in children of those mothers who did not do well in treatment. Donna Stewart, MD (Toronto, Canada), in her discussion, focused on the relationship between childhood abuse and later psychopathology. A general population study from Canada demonstrated that a history of child abuse increases all adult psychopathology but that there was a stronger effect in women than in men. She also noted that adversity, neglect, and abuse is a non-specific stressor for psychiatric disorder that has been shown to be associated with the later development of borderline personality disorder (BPD), substance abuse, antisocial PD, PTSD, somatization, anxiety and depression. Addressing the issue of why gender differences exist in anxiety and depression, she offered a number of possible explanations for this 2:1 difference in sex ratio, including that it may be an artifact of diagnostic bias or help-seeking, it may represents differences in socialization (women are encouraged to be nurturing and compliant), differences in cognitive style (woman are more ruminative), differences in social roles (women tend to assume more caring roles), differences in socio-economic status (more women are below the poverty line), woman are more likely to be victims of violence and abuse, discrimination at work and by society, and show differences in sex hormones genetics, and CNS functions. Taking the example of lone motherhood as a risk factor for GAD and MDD, 13% of all families in Canada are mother-led and make up 40% of all social assistance. The rate for MDD in the past 12 months was 19% for lone mothers vs.4.8% for mothers with partners.

Upon conclusion of the presentations, participants convened into two breakout groups to formulate recommendations for research and suggestions for future versions of the classifications of mental disorders (i.e., for DSM-V and ICD-11).

The first breakout group raised questions about what it means to say that two disorders are "closely related." The group opined being "closely-related" means that they are similar with regard to some but not all validators…like the relationship between bipolar disorder and schizophrenia. Neural substrates and pathophysiology would be important to establish but the data in this area is insufficient. Furthermore, there are discrepancies between different kinds of validators, for example. state vs. trait, family data vs. twin data, clinical vs. epidemiological samples, etc. The general consensus of the group was that MDD and GAD are related; most people in the group thought that they are closely related whereas some thought the closeness of their relationship was somewhere in the middle. The group then grappled with what were the implications of this, wondering how it could be justified to keep them separate in the DSM-V. Given that these categories were split from each other just a few iterations ago, can they legitimately be lumped again now, especially since lumping means you lose information you get from the diversity of categories? The group also considered the idea of a new overarching grouping for GAD and MDD and thought it would be important to come up with a non-stigmatizing name, like distress disorders or internalizing disorders, raising the question of what would be the underlying construct or organizing principle for this grouping; suggested possibilities included neuroticism or harm avoidance. The group also wondered what should be done about the somatoform disorders, which also appear to be associated with depression and anxiety and bipolar disorder. Even though bipolar disorder, like MDD, has high comorbidity with anxiety disorder, there is little associated between bipolar disorder and childhood adversity factors. The group agreed that the current definition of GAD in DSM-IV is lacking in many respects (should the duration be 2 weeks or 6 months? Should the anxiety be excessive or normal?). The group thought that it was hard to make a recommendation about the placement of GAD without a better definition. Regarding suggestions for additional data that is needed, the group strongly felt that there needs to be more of the biomarker data which is available for MDD. There also needs to be a better examination of mixed states-how prevalent and disabling are they and does there need to be a separate category for these mixed states?

The second breakout group began their discussion by focusing on terms of reference, noting that evidence that would strongly point to the extremes (i.e., that GAD and MDD are the same disorder or that MDD and GAD are completely different) would be hard to come by and that most evidence would likely point to the grey area in the middle. Most of the studies rejected these extremes and came down on the side of GAD and MDD being related disorders. For example, considering the genetic data, while twin studies suggest that they are the same disorder, family studies reject the idea that they are the same disorder but instead support the hypothesis that they are distally related. Biological evidence (imaging, endocrinology, neurotransmitters, and sleep findings) provide the most compelling evidence that they are distinct disorders. The group unanimously felt that GAD and MDD were closely related rather than being different forms of the same disorder or just distally related. The group also recommended that based on these findings at least some kind of change in the classification of MDD and GAD is warranted, a majority supported MDD and GAD being kept distinct but grouped together in the same diagnostic class to reflect the fact that they belong to a superordinate category that might be called "distress disorders." A minority in the group felt that the close relationship between MDD and GAD should be emphasized in the text descriptions of both disorders but that no other changes are justified. Regarding future research directions, the group recommended that individual items be re-examined with regard to onset and duration and subclinical symptoms, that secondary analyses be conducted to tease out drug effects, and that new research be commissioned into the underlying biology of GAD.

David Goldberg summarized the proceedings of the conference as follows:

There was general agreement that the two disorders were not identical, and that being "closely-related" means that they are similar with regard to some but not all validators. The view that they are distally related disorders was supported by a minority of those present. These took the view that there are discrepancies between different kinds of validators. There are inevitable differences in state measures of personality, and some family studies would favor the idea of distally related disorders. We had heard suggestive data from brain imaging, clinical and endocrinological data that these differences might be important. Since the categories were split from each other just a few iterations ago, can they legitimately be lumped again now, especially since lumping means you lose information you get from the diversity of categories? These members of the group took the view that while the close relationship between MDD and GAD should be emphasized in the text descriptions of both disorders in DSM-V, but that no other changes were justified.

Most of those present took the view that the two disorders are closely related, and took the view that the differences between the two were relatively trivial, and might be accounted for by the different time durations required for each. A disorder that has to last 6 months might be expected to have somewhat more pronounced vulnerability factors and somatic equivalents than one that only needed to last for 2 weeks. Those thinking they were closely related thought that some kind of change in the classification of MDD and GAD is warranted, with both disorders belonging to a superordinate category that might be given a non-stigmatizing name, like distress disorders or internalizing disorders. Those favoring "internalizing disorders" for the larger group including "fear disorders" suggested neuroticism or harm avoidance as the organizing principle, with GAD and MDE regarded as distress disorders.

In considering further steps that need to be taken in formulating recommendations for DSM-V, the high prevalence of mixed states and the duration to be required for each disorder need to be taken into account. Should such states need a separate category, or should anxious symptoms be considered to be common in depression, as in previous classifications? There is a need for further studies of the biomarker data which is available for MDD in patients with GAD.

Papers based on these presentations will appear in a forthcoming monograph to be published by American Psychiatric Publishing, Inc., which can be purchased here.


New User? Register Now
Forgot Password?

Related Links